The tests were conducted using a nice big box in calm fresh water. Cellphone technology, which 2.4GHz sets use, due to a bit of strange and unlikely maths that surfaced about 20 years ago, propogates with a spiral waveform, enabling it to reach into odd nooks and crannies better than a straight signal. An aerial mounted below waterline is relying on this, but the operative phrase is "reaches better", not "works infallibly".
Since microwaves, which also operate in this general band, work by having their energy absorbed by the liquid content of whatever they are heating, it should be no surprise that signals are lost in water. At cellphone powers, much the same as we use, the signal can penetrate 18mm of water before it is absorbed below usefulness. Reliance on the enhanced chance of a signal hitting the cavity formed in the water by the hull and finding the aerial rather than placing the aerial in line of sight is not a good pracitce, and will lead to poor reception and the problems that come with it.
Last week, a fellow club member lost signal in a fast boat on choppy salt water. The RX was mounted under the deck, but became dislodged and dropped into the bottom of the boat (about 1" below waterline when the boat stopped and settled) when between 50-100 yards out. Control was not re-established until the boat had drifted to within 10 yards of shore. The tests were an attempt at lab conditions, my mates boat was real life.
The difference between theory and practice is that in theory, there is no difference, in practice, there is.